- Email: [email protected]
Separate Time Behaviors of the Temporal and Frontal Mismatch Negativity Sources
NeuroImage 12, 14 –19 (2000) doi:10.1006/nimg.2000.0591, available online at http://www.idealibrary.com on
Separate Time Behaviors of the Temporal and Frontal Mismatch Negativity Sources T. Rinne,* K. Alho,* R. J. Ilmoniemi,† J. Virtanen,* ,† ,‡ and R. Na¨a¨ta¨nen,* ,† *Cognitive Brain Research Unit, Department of Psychology, University of Helsinki, Helsinki, Finland; †BioMag Laboratory, Medical Engineering Centre, and ‡Department of Radiology, Helsinki University Central Hospital, Helsinki, Finland Received June 24, 1999
scalp-potential recordings, which showed high amplitudes on a frontocentral electrode (Fz, frontal source) and on electrodes over the temporal lobe (T3 and T4, temporal source). Interestingly, the suggested source configuration was supported by later studies, which used more advanced source analysis tools. First, by using dipole modeling techniques, MMN (Scherg et al., 1989) and its magnetic counterpart (MMNm) (Hari et al., 1984) were found to be generated mainly in the auditory cortex in the temporal lobes. Second, the analysis of scalp-potential distribution suggested an additional, right-hemisphere MMN source, which supposedly was located in the frontal lobe (Giard et al., 1990; Deouell et al., 1998). Although MMN and the auditory cortex change-detection mechanism have ever since been objects of growing interest and the link between unattended auditory processing and attention is theoretically of great importance, surprisingly little is known about the frontal MMN generator and the significance of frontal lobes in the MMN process in general. The essential role of frontal areas in generating MMN is suggested by a recent study (Alain et al., 1998; see also Alho et al., 1994), which compared patients with focal lesions with normal subjects. It was shown that a unilateral prefrontal lesion diminishes the overall MMN amplitude, while a lesion in unilateral temporal cortex affects the amplitude only in the side of scalp contralateral to the stimulated ear. In addition, PET and fMRI studies have shown that areas in the frontal lobes, such as inferior frontal lobe and cingulate, are involved in the processing of auditory stimuli (Zatorre et al., 1992; Pugh et al., 1996; Celsis et al., 1999). Some of these areas might be involved in auditory target detection and participate in generating the frontal component of MMN. In the present study, the temporal behavior of the MMN and MMNm (the magnetic counterpart of MMN) source component structure is examined by using 64channel electroencephalography (EEG) and 122-channel magnetoencephalography (MEG) and up-to-date source analysis tools. Our aim was to explore whether the hypothesized time difference (Na¨a¨ta¨nen et al.,
It has been proposed that mismatch negativity (MMN) is generated by temporal and frontal lobe sources, the former being associated with change detection and the latter with involuntary switching of attention to sound change. If this switching of attention is triggered by the temporal cortex change-detection mechanism, one would expect that the frontal component of MMN is activated later than the temporal one. This was studied by using 64-channel electroencephalography (EEG) and 122-channel magnetoencephalography (MEG) with realistically shaped head models to determine the source current distribution in different lobes as a function of time. Minimum-norm estimation (MNE) was performed, constraining the solution to the reconstructed cortical sheet. The results support the hypothesis that the frontal MMN generator is activated later than the auditory cortex generator. © 2000 Academic Press
INTRODUCTION Mismatch negativity (MMN) (Na¨a¨ta¨nen et al., 1978; Na¨a¨ta¨nen, 1992) is a component of the scalp-recorded event-related potential (ERP) elicited by occasional changes in a sound sequence. MMN is elicited even when the subject is engaged in a demanding auditory or visual discrimination task and not paying attention to the stimuli. In the late seventies, when the MMN was first described, it was already hypothesized that the MMN mechanism forms the basis of unattended change detection, which may lead to the switching of attention. Further, it was suggested that this mechanism consists of an auditory cortex–frontal lobe network: “after the sensory analysis carried out by the sensory-specific systems has ‘detected’ a mismatch (reflected by the mismatch negativity over the respective areas), the frontal areas, known to play an important role in the elicitation of the orienting reflex, are activated” (Na¨a¨ta¨nen and Michie, 1979). This generator structure, i.e., a temporal and a frontal lobe MMN generator, was formulated on the basis of four-channel 1053-8119/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.
14
TIME BEHAVIORS OF MISMATCH NEGATIVITY SOURCES
1979; Giard et al., 1990) between temporal and frontal MMN components could be found. This would be an important step in establishing how the auditory cortex change-detection mechanism is connected to higherlevel processes, i.e., how involuntary attention switch is carried out by the brain. METHODS Subjects and procedure. Thirteen healthy righthanded adults volunteered as subjects (aged 19 –28, 7 females). The subjects were instructed to ignore the auditory stimuli while watching a silent self-selected movie on a monitor at the distance of about 2 m. Stimuli. The stimuli were spectrally rich tones, which consisted of three sinusoidal partials. The second and third components were 3 and 6 dB lower in intensity, respectively, than the first component. This tone structure was chosen to facilitate MMN elicitation: the MMN amplitude is larger with spectrally rich tones than with sinusoidal sounds (Na¨a¨ta¨nen et al., 1993; Tervaniemi et al., 1999, 2000). There were six different tones in the stimulus sequence: A frequent tone (P ⫽ 0.8) consisted of frequencies 500, 1000, and 1500 Hz and was 75 ms in duration (including 5-ms rise and fall times). The infrequent tones (P ⫽ 0.04 for each type) differed from the frequent tone in either duration (25 or 50 ms), intensity (15 dB lower), or frequency (⫾5 or ⫾10% change). The duration deviants elicited most replicable MMNs (Tervaniemi et al., 1999) and were therefore selected for further analysis reported here. The data for duration deviants are compared with the data for the other deviants elsewhere (Tervaniemi et al., 1999). The stimuli were delivered binaurally through plastic tubes and earpieces at an intensity of 60 dB above the hearing threshold determined separately for each subject. Frequency distortions of the tubes were compensated for with a correction filter. The constant stimulus onset-to-onset interval was 300 ms. The standard and deviant stimuli were delivered in random order, except that each deviant tone was preceded by at least two standard ones. Simultaneous EEG and MEG recording. The 64channel electroencephalogram was recorded (electrode on the tip of the nose as reference; Virtanen et al., 1996, 1997) simultaneously with whole-head 122-channel magnetoencephalogram (Neuromag Ltd.; passband 0.03–30 Hz, sampling frequency 250 Hz) (Fig. 1). The position of the subject’s head relative to the magnetometer was determined by measuring the magnetic field produced by three marker coils attached to the electrode cap. The three-dimensional (3D) locations of the electrodes and the marker coils in relation to cardinal points on the head were measured with an Isotrack 3D digitizer (Polhemus, USA). Horizontal and vertical eye movements were monitored with separate bipolar elec-
15
trooculogram (EOG) electrodes placed above and below the left eye and lateral to the eyes. The recordings were conducted in a magnetically shielded room. The EEG and MEG epochs, time-locked to the stimuli starting 100 ms before (prestimulus baseline) and ending 300 ms after stimulus onset, were averaged separately for standard and deviant stimuli. Epochs with deflections larger than 150 V at any of the EOG or EEG channels were rejected from averaging because they were probably caused by extracerebral artifacts such as eye movements, blinks, and muscle activity. Recording was continued until at least 500 accepted deviant responses were collected or the session had lasted 60 min. On average, a recording session lasted 55 min. Data reduction. First, the responses to the frequent 75-ms tones were subtracted from the responses to the infrequent 25-ms tones to reveal the MMN and MMNm. Then, for each subject, minimum-norm estimation (MNE; Curry software, Philips GmbH, Germany) was performed to estimate the MMN/MMNm source-current distribution as a function of time. The MNE solution was constrained to the reconstructed cortical sheet. For each time instant, the squared sum of the discretized MNE amplitudes was calculated separately for frontal and temporal activation in each hemisphere. These two activations were separated by a plane tilted approximately along the Sylvian fissures in both hemispheres. Finally, the latency of the maximum activation was determined (no interpolation) for each hemisphere and lobe. Magnetic resonance images. A 3D set of magnetic resonance images (MRIs; Siemens Vision 1.5-T system) of the head was obtained for each subject. The skin and cortical surfaces for the head model and visualization were segmented with the Curry software. Electrodes were overlaid with respect to the digitized anatomical landmarks, i.e., preauricular points and the nasion. On the basis of the individual MRI, realistically shaped boundary element head models (approximately 3000 nodes) (Fig. 1) were determined for each subject. RESULTS AND DISCUSSION All 13 subjects showed a prominent MMN and MMNm in response to the infrequent 25-ms tones presented among the frequent 75-ms tones. A typical MMN waveform and scalp-potential distribution are illustrated in Figs. 2 and 3 (top), respectively. The potential distribution closely corresponds to one that would be generated by two dipolar sources located in the temporal lobes of each hemisphere. However, minimum-norm estimate (MNE) source current maps (Fig. 3, middle) calculated from these scalp potentials (Fig. 3, top) reveal a somewhat different spatiotemporal source pattern. At 160 ms, the maximum of activation is seen over the auditory cortex of the temporal lobe. As a function of time, the center of gravity of the esti-
16
RINNE ET AL.
FIG. 1. Left: Location of the 64-channel EEG (circles) and 122-channel MEG (two orthogonal planar gradiometers in each square) sensor arrays with respect to the head. Right: Realistically shaped head model surfaces based on the individual MRI.
mated source current distribution moves anteriorly, indicating activation of an additional source (or sources) with a more frontal location. Interestingly, in the MEG-based MNE map (Fig. 3, bottom), activation can be seen over the temporal cortex only and no later frontal activation is present as in the EEG-based map. This difference between EEG and MEG probably reflects the nature of the source generating the frontal EEG activation. MEG signal arises mainly from sources oriented tangentially with respect to the scalp. In addition, deep sources are hard to detect with MEG. Thus, the frontal MMN activation is probably generated either in superficial cortical areas, where the
sources are radial, or deep in the brain. These differences between the lead fields of EEG and MEG, i.e., in their sensitivity patterns, probably explain the difference between the EEG-based (Fig. 3, middle) and MEG-based (Fig. 3, bottom) estimates of the current distributions. Figure 4 shows the time behavior of the right-hemisphere frontal and temporal activations extracted from the MNE map (Fig. 3, middle). The temporal activation peaks at 160 ms while the maximum of frontal activation is reached later, at 168 ms. The time difference between the temporal and frontal peak activations seems to be smaller in the curves (Fig. 4) than in the MNE map (Fig. 3, middle), for the sum of squares of the MNE items is affected by a large area (red in the MNE map) around the peak (yellow in the MNE map). The peak latencies for all subjects are presented in Fig. 5. The time difference between frontal and temporal activation was significant only in the right-hemisphere EEG data. On average, the right-hemisphere frontal MMN activation peaked later than the temporal activation (Friedman’s nonparametric ANOVA, P ⬍ 0.01). The mean difference was about 8 ms. It is important to note how the data analysis is affected by spatial smearing. The frontal and temporal source activations (Fig. 4) cannot be spatially fully separated from one other. Because of this, it is not possible to compare, for example, activation onsets of the two sources as, at the onset, both source curves are most likely reflecting the activation of the dominating temporal source. The source activation curves (Fig. 4) provide meaningful
FIG. 2. Mean global field power (left) and waveforms from individual EEG and MEG sensors (right) illustrating the MMN strength as a function of time in Subject I.J. The illustrated data are obtained by subtracting responses to frequent stimuli from responses to infrequent stimuli. MMN is peaking at about 160 ms from stimulus onset. The three latencies shown in subsequent figures are marked with vertical lines (left). EEG data (upper right) are shown from a frontal midline electrode (common-average reference) and the MEG sensor (lower right) is located over the right auditory cortex.
FIG. 3. Top: Isopotential maps (increment 0.5 V, common average reference) showing the scalp distribution of the MMN in Subject I.J. at different latencies. The MMN derives its name from the frontocentral negative maximum (negativity marked in yellow). The maps show a two-dimensional projection of the potential distribution seen from above. The electrode locations are marked with grey circles. Middle: MNE map calculated from the scalp potential. At 160 ms from stimulus onset, the activation shows a temporal maximum (yellow), indicating an auditory cortex source. As a function of time, the center of gravity of activation moves to a more frontal location. Bottom: MNE map calculated from the magnetic field. For this subject, a temporal maximum, indicating auditory cortex MMNm source, can be seen, but no later frontal activation as in the EEG based MNE. 17
18
RINNE ET AL.
FIG. 4. Time course of the temporal and frontal MMN activation in the right hemisphere calculated from the EEG-based MNE map (Fig. 3, middle). The chart presents the sum of the squares of the MNE items for each time point in temporal and frontal areas.
information only if there are time differences in the peak amplitudes. The results thus support the hypothesis that some frontal areas are activated during the MMN response and that these frontal areas are activated after the temporal auditory cortex source. Previously, it had been shown that the scalp distribution of the MMN evoked by changes in simple tones is right-hemisphere predominant (Giard et al., 1990; Paavilainen et al., 1991; Leva¨nen et al., 1996) and that this predominance might be caused by an additional source probably located in the frontal lobe (Giard et al., 1990; Deouell et al., 1998). In addition, indirect support for frontal lobe involvement in the MMN process is provided by studies that showed diminished MMN amplitudes in patients with a prefrontal cortex lesion, while the auditory cortex-generated N1 response was unaffected (Alho et al., 1994; Alain et al., 1998) and, further, that a unilateral prefrontal lesion diminished the overall MMN amplitude, whereas a unilateral temporal cortex lesion affected the amplitude only on the side of scalp contralateral to the stimulation (Alain et al., 1998). It is plausible that the frontal component of the MMN is generated by some brain mechanism related to involuntary attention switch as it has been shown that small changes in task-irrelevant auditory stimuli elicit MMN even when the subject is performing an attention-demanding task and, further, that these small unattended changes deteriorate the subject’s performance in the task (Schro¨ger, 1996; Alho et al., 1997; Escera et al., 1998). Taken together, the present and previous data support the existence of an auditory cortex–frontal lobe network responsible for an involuntary attention switch to changes in the auditory environment. The present study demonstrated that the frontal MMN activation can be measured on the level of individual subjects and separated in time from the audi-
tory cortex source. Although the time difference between the temporal and frontal activation was significant only in the right hemisphere, some subjects showed a similar pattern also in the left hemisphere (Fig. 5), which probably reflects individual variation in brain structures and in the MMN source configuration. We demonstrated how the excellent time resolution of EEG can be exploited to follow changes in brain activation within a few milliseconds. To achieve this goal, source-analysis tools must be used, for the source level information is not directly seen in the scalp-recorded waveforms. It is noteworthy that different source-analysis techniques attain different aspects of brain function. For example, dipole modeling might not be the ideal choice to extract the spatiotemporal source structure of the MMN. This is because the auditory cortex activation is dominating the MMN field pattern, rendering it difficult to tell apart additional sources, which might be smaller in amplitude, deeper, or distributed. By using MNE, the spatiotemporal structure of MMN sources can be visualized. CONCLUSIONS The present data provide evidence for the existence of a frontal MMN generator probably underlying atten-
FIG. 5. EEG (top) and MEG (bottom) peak temporal and frontal activation for each subject in both hemispheres. In the right hemisphere, the frontal MMN activation peaked later than the temporal MMN activation (top right). Note that equal peak latencies for temporal and frontal activation indicate that the frontal activation cannot be separated from the dominating temporal activation. This might be because of low signal-to-noise ratio or because of the orientation of the frontal source.
TIME BEHAVIORS OF MISMATCH NEGATIVITY SOURCES
tion switch to a change in repetitive auditory stimuli. The time difference between the frontal and temporal peak activations supports the hypothesis that the frontal MMN component is initiated by the auditory cortex change-detection mechanism and could be connected to a mechanism causing an involuntary switch of attention. The fact that no frontal activity was seen with MEG suggests that the source of the frontal MMN component either is located deeper in the brain or is radially oriented as these kinds of sources are not well detected by MEG (Ha¨ma¨la¨inen et al., 1993). However, the exact location of the frontal source is not known but on the basis of auditory fMRI and PET studies the possible areas include inferior frontal cortex, superior frontal gyrus, and cingulate. Thus, the frontal MMN component might actually be generated by a distributed network of sources. ACKNOWLEDGMENTS This study was supported by EV contract BMH4-CT-0819 (COBRAIN) and by the Academy of Finland. The authors thank Antti Pitka¨nen and Markus Holi for their help in conducting the experiments and in data analysis.
REFERENCES Alain, C., Woods, D. L., and Knight, R. T. 1998. A distributed cortical network for auditory sensory memory in humans. Brain Res. 812: 23–37. Alho, K., Escera, C., Diaz, R., Yago, E., and Serra, J. M. 1997. Effects of involuntary auditory attention on visual task performance and brain activity. NeuroReport 8:3233–3237. Alho, K., Woods, D. L., Algazi, A., Knight, R. T., and Na¨a¨ta¨nen, R. 1994. Lesions of frontal cortex diminish the auditory mismatch negativity. Electroencephalogr. Clin. Neurophysiol. 91:353–362. Celsis, P., Boulanouar, K., Doyon, B., Ranjeva, J. P., Berry, I., Nespoulous, J. L., and Chollet, F. 1999. Differential fMRI responses in the left posterior superior temporal gyrus and left supramarginal gyrus to habituation and change detection in syllables and tones. NeuroImage 9:135–144. Deouell, L. Y., Bentin, S., and Giard, M. H. 1998. Mismatch negativity in dichotic listening: Evidence for interhemispheric differences and multiple generators. Psychophysiology 35:355–365. Escera, C., Alho, K., Winkler, I., and Na¨a¨ta¨nen, R. 1998. Neural mechanisms of involuntary attention to acoustic novelty and change. J. Cogn. Neurosci. 10:590 – 604. Giard, M. H., Perrin, F., Pernier, J., and Bouchet, P. 1990. Brain generators implicated in the processing of auditory stimulus deviance: A topographic event-related potential study. Psychophysiology 27:627– 640. Ha¨ma¨la¨inen, M., Hari, R., Ilmoniemi, R. J., Knuutila, J., and Lounasmaa, O. V. 1993. Magnetoencephalography-theory, instrumen-
19
tation, and applications to noinvasive studies of the working human brain. Rev. Mod. Phys. 65:413– 497. Hari, R., Ha¨ma¨la¨inen, M., Ilmoniemi, R., Kaukoranta, E., Reinikainen, K., Salminen, J., Alho, K., Na¨a¨ta¨nen, R., and Sams, M. 1984. Responses of the primary auditory cortex to pitch changes in a sequence of tone pips: Neuromagnetic recordings in man. Neurosci. Lett. 50:127–132. Leva¨nen, S., Ahonen, A., Hari, R., McEvoy, L., and Sams, M. 1996. Deviant auditory stimuli activate human left and right auditory cortex differently. Cereb. Cortex 6:288 –296. Na¨a¨ta¨nen, R. 1992. Attention and Brain Function. Lawrence Erlbaum, Hillsdale, NJ. Na¨a¨ta¨nen, R., and Michie, P. T. 1979. Early selective-attention effects on the evoked potential: A critical review and reinterpretation. Biol. Psychol. 8:81–136. Na¨a¨ta¨nen, R., Schro¨ger, E., Karakas, S., Tervaniemi, M., and Paavilainen, P. 1993. Development of a memory trace for a complex sound in the human brain. NeuroReport 4:503–506. Na¨a¨ta¨nen, R., Gaillard, A. W. K., and Ma¨ntysalo, S. 1978. Early selective-attention effect on evoked potential reinterpreted. Acta Psychol. 42:313–329. Paavilainen, P., Alho, K., Reinikainen, K., Sams, M., and Na¨a¨ta¨nen, R. 1991. Right hemisphere dominance of different mismatch negativities. Electroencephalogr. Clin. Neurophysiol. 78:466 – 479. Pugh, K. R., Shaywitz, B. A., Shaywitz, S. E., Fulbright, R. K., Byrd, D., Skudlarski, P., Shankweiler, D. P., Katz, L., Constable, R. T., Fletcher, J., Lacadie, C., Marchione, K., and Gore, J. C. 1996. Auditory selective attention: An fMRI investigation. NeuroImage 4:159 –173. Scherg, M., Vajsar, J., and Picton, T. W. 1989. A source analysis of the late human auditory evoked potentials. J. Cogn. Neurosci. 1:336 –355. Schro¨ger, E. 1996. A neural mechanism for involuntary attention shifts to changes in auditory stimulation. J. Cogn. Neurosci. 8:527–539. Tervaniemi, M., Ilvonen, T., Sinkkonen, J., Kujala, A., Alho, K., Huotilainen, M., and Na¨a¨ta¨nen, R. 2000. Harmonic partials facilitate pitch discrimination in humans: Electrophysiological and behavioral evidence. Neurosci. Lett. 279:29 –32. Tervaniemi, M., Lehtokoski, A., Sinkkonen, J., Virtanen, J., Ilmoniemi, R. J., and Na¨a¨ta¨nen, R. 1999. Test–retest reliability of mismatch negativity for duration, frequency, and intensity changes. Clin. Neurophysiol. 110:1388 –1393. Virtanen, J., Parkkonen, L., Ilmoniemi, R. J., Pekkonen, E., and Na¨a¨ta¨nen, R. 1997. Biopotential amplifier for simultaneous operation with biomagnetic instruments. Med. Biol. Eng. Comp. 35: 402– 408. Virtanen, J., Rinne, T., Ilmoniemi, R. J., and Na¨a¨ta¨nen, R. 1996. MEG-compatible multichannel EEG electrode array. Electroencephalogr. Clin. Neurophysiol. 99:568 –570. Zatorre, R. J., Evans, A. C., Meyer, E., and Gjedde, A. 1992. Lateralization of phonetic and pitch discrimination in speech processing. Science 256:846 – 849.
Separate Time Behaviors of the Temporal and Frontal Mismatch Negativity Sources T. Rinne,* K. Alho,* R. J. Ilmoniemi,† J. Virtanen,* ,† ,‡ and R. Na¨a¨ta¨nen,* ,† *Cognitive Brain Research Unit, Department of Psychology, University of Helsinki, Helsinki, Finland; †BioMag Laboratory, Medical Engineering Centre, and ‡Department of Radiology, Helsinki University Central Hospital, Helsinki, Finland Received June 24, 1999
scalp-potential recordings, which showed high amplitudes on a frontocentral electrode (Fz, frontal source) and on electrodes over the temporal lobe (T3 and T4, temporal source). Interestingly, the suggested source configuration was supported by later studies, which used more advanced source analysis tools. First, by using dipole modeling techniques, MMN (Scherg et al., 1989) and its magnetic counterpart (MMNm) (Hari et al., 1984) were found to be generated mainly in the auditory cortex in the temporal lobes. Second, the analysis of scalp-potential distribution suggested an additional, right-hemisphere MMN source, which supposedly was located in the frontal lobe (Giard et al., 1990; Deouell et al., 1998). Although MMN and the auditory cortex change-detection mechanism have ever since been objects of growing interest and the link between unattended auditory processing and attention is theoretically of great importance, surprisingly little is known about the frontal MMN generator and the significance of frontal lobes in the MMN process in general. The essential role of frontal areas in generating MMN is suggested by a recent study (Alain et al., 1998; see also Alho et al., 1994), which compared patients with focal lesions with normal subjects. It was shown that a unilateral prefrontal lesion diminishes the overall MMN amplitude, while a lesion in unilateral temporal cortex affects the amplitude only in the side of scalp contralateral to the stimulated ear. In addition, PET and fMRI studies have shown that areas in the frontal lobes, such as inferior frontal lobe and cingulate, are involved in the processing of auditory stimuli (Zatorre et al., 1992; Pugh et al., 1996; Celsis et al., 1999). Some of these areas might be involved in auditory target detection and participate in generating the frontal component of MMN. In the present study, the temporal behavior of the MMN and MMNm (the magnetic counterpart of MMN) source component structure is examined by using 64channel electroencephalography (EEG) and 122-channel magnetoencephalography (MEG) and up-to-date source analysis tools. Our aim was to explore whether the hypothesized time difference (Na¨a¨ta¨nen et al.,
It has been proposed that mismatch negativity (MMN) is generated by temporal and frontal lobe sources, the former being associated with change detection and the latter with involuntary switching of attention to sound change. If this switching of attention is triggered by the temporal cortex change-detection mechanism, one would expect that the frontal component of MMN is activated later than the temporal one. This was studied by using 64-channel electroencephalography (EEG) and 122-channel magnetoencephalography (MEG) with realistically shaped head models to determine the source current distribution in different lobes as a function of time. Minimum-norm estimation (MNE) was performed, constraining the solution to the reconstructed cortical sheet. The results support the hypothesis that the frontal MMN generator is activated later than the auditory cortex generator. © 2000 Academic Press
INTRODUCTION Mismatch negativity (MMN) (Na¨a¨ta¨nen et al., 1978; Na¨a¨ta¨nen, 1992) is a component of the scalp-recorded event-related potential (ERP) elicited by occasional changes in a sound sequence. MMN is elicited even when the subject is engaged in a demanding auditory or visual discrimination task and not paying attention to the stimuli. In the late seventies, when the MMN was first described, it was already hypothesized that the MMN mechanism forms the basis of unattended change detection, which may lead to the switching of attention. Further, it was suggested that this mechanism consists of an auditory cortex–frontal lobe network: “after the sensory analysis carried out by the sensory-specific systems has ‘detected’ a mismatch (reflected by the mismatch negativity over the respective areas), the frontal areas, known to play an important role in the elicitation of the orienting reflex, are activated” (Na¨a¨ta¨nen and Michie, 1979). This generator structure, i.e., a temporal and a frontal lobe MMN generator, was formulated on the basis of four-channel 1053-8119/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.
14
TIME BEHAVIORS OF MISMATCH NEGATIVITY SOURCES
1979; Giard et al., 1990) between temporal and frontal MMN components could be found. This would be an important step in establishing how the auditory cortex change-detection mechanism is connected to higherlevel processes, i.e., how involuntary attention switch is carried out by the brain. METHODS Subjects and procedure. Thirteen healthy righthanded adults volunteered as subjects (aged 19 –28, 7 females). The subjects were instructed to ignore the auditory stimuli while watching a silent self-selected movie on a monitor at the distance of about 2 m. Stimuli. The stimuli were spectrally rich tones, which consisted of three sinusoidal partials. The second and third components were 3 and 6 dB lower in intensity, respectively, than the first component. This tone structure was chosen to facilitate MMN elicitation: the MMN amplitude is larger with spectrally rich tones than with sinusoidal sounds (Na¨a¨ta¨nen et al., 1993; Tervaniemi et al., 1999, 2000). There were six different tones in the stimulus sequence: A frequent tone (P ⫽ 0.8) consisted of frequencies 500, 1000, and 1500 Hz and was 75 ms in duration (including 5-ms rise and fall times). The infrequent tones (P ⫽ 0.04 for each type) differed from the frequent tone in either duration (25 or 50 ms), intensity (15 dB lower), or frequency (⫾5 or ⫾10% change). The duration deviants elicited most replicable MMNs (Tervaniemi et al., 1999) and were therefore selected for further analysis reported here. The data for duration deviants are compared with the data for the other deviants elsewhere (Tervaniemi et al., 1999). The stimuli were delivered binaurally through plastic tubes and earpieces at an intensity of 60 dB above the hearing threshold determined separately for each subject. Frequency distortions of the tubes were compensated for with a correction filter. The constant stimulus onset-to-onset interval was 300 ms. The standard and deviant stimuli were delivered in random order, except that each deviant tone was preceded by at least two standard ones. Simultaneous EEG and MEG recording. The 64channel electroencephalogram was recorded (electrode on the tip of the nose as reference; Virtanen et al., 1996, 1997) simultaneously with whole-head 122-channel magnetoencephalogram (Neuromag Ltd.; passband 0.03–30 Hz, sampling frequency 250 Hz) (Fig. 1). The position of the subject’s head relative to the magnetometer was determined by measuring the magnetic field produced by three marker coils attached to the electrode cap. The three-dimensional (3D) locations of the electrodes and the marker coils in relation to cardinal points on the head were measured with an Isotrack 3D digitizer (Polhemus, USA). Horizontal and vertical eye movements were monitored with separate bipolar elec-
15
trooculogram (EOG) electrodes placed above and below the left eye and lateral to the eyes. The recordings were conducted in a magnetically shielded room. The EEG and MEG epochs, time-locked to the stimuli starting 100 ms before (prestimulus baseline) and ending 300 ms after stimulus onset, were averaged separately for standard and deviant stimuli. Epochs with deflections larger than 150 V at any of the EOG or EEG channels were rejected from averaging because they were probably caused by extracerebral artifacts such as eye movements, blinks, and muscle activity. Recording was continued until at least 500 accepted deviant responses were collected or the session had lasted 60 min. On average, a recording session lasted 55 min. Data reduction. First, the responses to the frequent 75-ms tones were subtracted from the responses to the infrequent 25-ms tones to reveal the MMN and MMNm. Then, for each subject, minimum-norm estimation (MNE; Curry software, Philips GmbH, Germany) was performed to estimate the MMN/MMNm source-current distribution as a function of time. The MNE solution was constrained to the reconstructed cortical sheet. For each time instant, the squared sum of the discretized MNE amplitudes was calculated separately for frontal and temporal activation in each hemisphere. These two activations were separated by a plane tilted approximately along the Sylvian fissures in both hemispheres. Finally, the latency of the maximum activation was determined (no interpolation) for each hemisphere and lobe. Magnetic resonance images. A 3D set of magnetic resonance images (MRIs; Siemens Vision 1.5-T system) of the head was obtained for each subject. The skin and cortical surfaces for the head model and visualization were segmented with the Curry software. Electrodes were overlaid with respect to the digitized anatomical landmarks, i.e., preauricular points and the nasion. On the basis of the individual MRI, realistically shaped boundary element head models (approximately 3000 nodes) (Fig. 1) were determined for each subject. RESULTS AND DISCUSSION All 13 subjects showed a prominent MMN and MMNm in response to the infrequent 25-ms tones presented among the frequent 75-ms tones. A typical MMN waveform and scalp-potential distribution are illustrated in Figs. 2 and 3 (top), respectively. The potential distribution closely corresponds to one that would be generated by two dipolar sources located in the temporal lobes of each hemisphere. However, minimum-norm estimate (MNE) source current maps (Fig. 3, middle) calculated from these scalp potentials (Fig. 3, top) reveal a somewhat different spatiotemporal source pattern. At 160 ms, the maximum of activation is seen over the auditory cortex of the temporal lobe. As a function of time, the center of gravity of the esti-
16
RINNE ET AL.
FIG. 1. Left: Location of the 64-channel EEG (circles) and 122-channel MEG (two orthogonal planar gradiometers in each square) sensor arrays with respect to the head. Right: Realistically shaped head model surfaces based on the individual MRI.
mated source current distribution moves anteriorly, indicating activation of an additional source (or sources) with a more frontal location. Interestingly, in the MEG-based MNE map (Fig. 3, bottom), activation can be seen over the temporal cortex only and no later frontal activation is present as in the EEG-based map. This difference between EEG and MEG probably reflects the nature of the source generating the frontal EEG activation. MEG signal arises mainly from sources oriented tangentially with respect to the scalp. In addition, deep sources are hard to detect with MEG. Thus, the frontal MMN activation is probably generated either in superficial cortical areas, where the
sources are radial, or deep in the brain. These differences between the lead fields of EEG and MEG, i.e., in their sensitivity patterns, probably explain the difference between the EEG-based (Fig. 3, middle) and MEG-based (Fig. 3, bottom) estimates of the current distributions. Figure 4 shows the time behavior of the right-hemisphere frontal and temporal activations extracted from the MNE map (Fig. 3, middle). The temporal activation peaks at 160 ms while the maximum of frontal activation is reached later, at 168 ms. The time difference between the temporal and frontal peak activations seems to be smaller in the curves (Fig. 4) than in the MNE map (Fig. 3, middle), for the sum of squares of the MNE items is affected by a large area (red in the MNE map) around the peak (yellow in the MNE map). The peak latencies for all subjects are presented in Fig. 5. The time difference between frontal and temporal activation was significant only in the right-hemisphere EEG data. On average, the right-hemisphere frontal MMN activation peaked later than the temporal activation (Friedman’s nonparametric ANOVA, P ⬍ 0.01). The mean difference was about 8 ms. It is important to note how the data analysis is affected by spatial smearing. The frontal and temporal source activations (Fig. 4) cannot be spatially fully separated from one other. Because of this, it is not possible to compare, for example, activation onsets of the two sources as, at the onset, both source curves are most likely reflecting the activation of the dominating temporal source. The source activation curves (Fig. 4) provide meaningful
FIG. 2. Mean global field power (left) and waveforms from individual EEG and MEG sensors (right) illustrating the MMN strength as a function of time in Subject I.J. The illustrated data are obtained by subtracting responses to frequent stimuli from responses to infrequent stimuli. MMN is peaking at about 160 ms from stimulus onset. The three latencies shown in subsequent figures are marked with vertical lines (left). EEG data (upper right) are shown from a frontal midline electrode (common-average reference) and the MEG sensor (lower right) is located over the right auditory cortex.
FIG. 3. Top: Isopotential maps (increment 0.5 V, common average reference) showing the scalp distribution of the MMN in Subject I.J. at different latencies. The MMN derives its name from the frontocentral negative maximum (negativity marked in yellow). The maps show a two-dimensional projection of the potential distribution seen from above. The electrode locations are marked with grey circles. Middle: MNE map calculated from the scalp potential. At 160 ms from stimulus onset, the activation shows a temporal maximum (yellow), indicating an auditory cortex source. As a function of time, the center of gravity of activation moves to a more frontal location. Bottom: MNE map calculated from the magnetic field. For this subject, a temporal maximum, indicating auditory cortex MMNm source, can be seen, but no later frontal activation as in the EEG based MNE. 17
18
RINNE ET AL.
FIG. 4. Time course of the temporal and frontal MMN activation in the right hemisphere calculated from the EEG-based MNE map (Fig. 3, middle). The chart presents the sum of the squares of the MNE items for each time point in temporal and frontal areas.
information only if there are time differences in the peak amplitudes. The results thus support the hypothesis that some frontal areas are activated during the MMN response and that these frontal areas are activated after the temporal auditory cortex source. Previously, it had been shown that the scalp distribution of the MMN evoked by changes in simple tones is right-hemisphere predominant (Giard et al., 1990; Paavilainen et al., 1991; Leva¨nen et al., 1996) and that this predominance might be caused by an additional source probably located in the frontal lobe (Giard et al., 1990; Deouell et al., 1998). In addition, indirect support for frontal lobe involvement in the MMN process is provided by studies that showed diminished MMN amplitudes in patients with a prefrontal cortex lesion, while the auditory cortex-generated N1 response was unaffected (Alho et al., 1994; Alain et al., 1998) and, further, that a unilateral prefrontal lesion diminished the overall MMN amplitude, whereas a unilateral temporal cortex lesion affected the amplitude only on the side of scalp contralateral to the stimulation (Alain et al., 1998). It is plausible that the frontal component of the MMN is generated by some brain mechanism related to involuntary attention switch as it has been shown that small changes in task-irrelevant auditory stimuli elicit MMN even when the subject is performing an attention-demanding task and, further, that these small unattended changes deteriorate the subject’s performance in the task (Schro¨ger, 1996; Alho et al., 1997; Escera et al., 1998). Taken together, the present and previous data support the existence of an auditory cortex–frontal lobe network responsible for an involuntary attention switch to changes in the auditory environment. The present study demonstrated that the frontal MMN activation can be measured on the level of individual subjects and separated in time from the audi-
tory cortex source. Although the time difference between the temporal and frontal activation was significant only in the right hemisphere, some subjects showed a similar pattern also in the left hemisphere (Fig. 5), which probably reflects individual variation in brain structures and in the MMN source configuration. We demonstrated how the excellent time resolution of EEG can be exploited to follow changes in brain activation within a few milliseconds. To achieve this goal, source-analysis tools must be used, for the source level information is not directly seen in the scalp-recorded waveforms. It is noteworthy that different source-analysis techniques attain different aspects of brain function. For example, dipole modeling might not be the ideal choice to extract the spatiotemporal source structure of the MMN. This is because the auditory cortex activation is dominating the MMN field pattern, rendering it difficult to tell apart additional sources, which might be smaller in amplitude, deeper, or distributed. By using MNE, the spatiotemporal structure of MMN sources can be visualized. CONCLUSIONS The present data provide evidence for the existence of a frontal MMN generator probably underlying atten-
FIG. 5. EEG (top) and MEG (bottom) peak temporal and frontal activation for each subject in both hemispheres. In the right hemisphere, the frontal MMN activation peaked later than the temporal MMN activation (top right). Note that equal peak latencies for temporal and frontal activation indicate that the frontal activation cannot be separated from the dominating temporal activation. This might be because of low signal-to-noise ratio or because of the orientation of the frontal source.
TIME BEHAVIORS OF MISMATCH NEGATIVITY SOURCES
tion switch to a change in repetitive auditory stimuli. The time difference between the frontal and temporal peak activations supports the hypothesis that the frontal MMN component is initiated by the auditory cortex change-detection mechanism and could be connected to a mechanism causing an involuntary switch of attention. The fact that no frontal activity was seen with MEG suggests that the source of the frontal MMN component either is located deeper in the brain or is radially oriented as these kinds of sources are not well detected by MEG (Ha¨ma¨la¨inen et al., 1993). However, the exact location of the frontal source is not known but on the basis of auditory fMRI and PET studies the possible areas include inferior frontal cortex, superior frontal gyrus, and cingulate. Thus, the frontal MMN component might actually be generated by a distributed network of sources. ACKNOWLEDGMENTS This study was supported by EV contract BMH4-CT-0819 (COBRAIN) and by the Academy of Finland. The authors thank Antti Pitka¨nen and Markus Holi for their help in conducting the experiments and in data analysis.
REFERENCES Alain, C., Woods, D. L., and Knight, R. T. 1998. A distributed cortical network for auditory sensory memory in humans. Brain Res. 812: 23–37. Alho, K., Escera, C., Diaz, R., Yago, E., and Serra, J. M. 1997. Effects of involuntary auditory attention on visual task performance and brain activity. NeuroReport 8:3233–3237. Alho, K., Woods, D. L., Algazi, A., Knight, R. T., and Na¨a¨ta¨nen, R. 1994. Lesions of frontal cortex diminish the auditory mismatch negativity. Electroencephalogr. Clin. Neurophysiol. 91:353–362. Celsis, P., Boulanouar, K., Doyon, B., Ranjeva, J. P., Berry, I., Nespoulous, J. L., and Chollet, F. 1999. Differential fMRI responses in the left posterior superior temporal gyrus and left supramarginal gyrus to habituation and change detection in syllables and tones. NeuroImage 9:135–144. Deouell, L. Y., Bentin, S., and Giard, M. H. 1998. Mismatch negativity in dichotic listening: Evidence for interhemispheric differences and multiple generators. Psychophysiology 35:355–365. Escera, C., Alho, K., Winkler, I., and Na¨a¨ta¨nen, R. 1998. Neural mechanisms of involuntary attention to acoustic novelty and change. J. Cogn. Neurosci. 10:590 – 604. Giard, M. H., Perrin, F., Pernier, J., and Bouchet, P. 1990. Brain generators implicated in the processing of auditory stimulus deviance: A topographic event-related potential study. Psychophysiology 27:627– 640. Ha¨ma¨la¨inen, M., Hari, R., Ilmoniemi, R. J., Knuutila, J., and Lounasmaa, O. V. 1993. Magnetoencephalography-theory, instrumen-
19
tation, and applications to noinvasive studies of the working human brain. Rev. Mod. Phys. 65:413– 497. Hari, R., Ha¨ma¨la¨inen, M., Ilmoniemi, R., Kaukoranta, E., Reinikainen, K., Salminen, J., Alho, K., Na¨a¨ta¨nen, R., and Sams, M. 1984. Responses of the primary auditory cortex to pitch changes in a sequence of tone pips: Neuromagnetic recordings in man. Neurosci. Lett. 50:127–132. Leva¨nen, S., Ahonen, A., Hari, R., McEvoy, L., and Sams, M. 1996. Deviant auditory stimuli activate human left and right auditory cortex differently. Cereb. Cortex 6:288 –296. Na¨a¨ta¨nen, R. 1992. Attention and Brain Function. Lawrence Erlbaum, Hillsdale, NJ. Na¨a¨ta¨nen, R., and Michie, P. T. 1979. Early selective-attention effects on the evoked potential: A critical review and reinterpretation. Biol. Psychol. 8:81–136. Na¨a¨ta¨nen, R., Schro¨ger, E., Karakas, S., Tervaniemi, M., and Paavilainen, P. 1993. Development of a memory trace for a complex sound in the human brain. NeuroReport 4:503–506. Na¨a¨ta¨nen, R., Gaillard, A. W. K., and Ma¨ntysalo, S. 1978. Early selective-attention effect on evoked potential reinterpreted. Acta Psychol. 42:313–329. Paavilainen, P., Alho, K., Reinikainen, K., Sams, M., and Na¨a¨ta¨nen, R. 1991. Right hemisphere dominance of different mismatch negativities. Electroencephalogr. Clin. Neurophysiol. 78:466 – 479. Pugh, K. R., Shaywitz, B. A., Shaywitz, S. E., Fulbright, R. K., Byrd, D., Skudlarski, P., Shankweiler, D. P., Katz, L., Constable, R. T., Fletcher, J., Lacadie, C., Marchione, K., and Gore, J. C. 1996. Auditory selective attention: An fMRI investigation. NeuroImage 4:159 –173. Scherg, M., Vajsar, J., and Picton, T. W. 1989. A source analysis of the late human auditory evoked potentials. J. Cogn. Neurosci. 1:336 –355. Schro¨ger, E. 1996. A neural mechanism for involuntary attention shifts to changes in auditory stimulation. J. Cogn. Neurosci. 8:527–539. Tervaniemi, M., Ilvonen, T., Sinkkonen, J., Kujala, A., Alho, K., Huotilainen, M., and Na¨a¨ta¨nen, R. 2000. Harmonic partials facilitate pitch discrimination in humans: Electrophysiological and behavioral evidence. Neurosci. Lett. 279:29 –32. Tervaniemi, M., Lehtokoski, A., Sinkkonen, J., Virtanen, J., Ilmoniemi, R. J., and Na¨a¨ta¨nen, R. 1999. Test–retest reliability of mismatch negativity for duration, frequency, and intensity changes. Clin. Neurophysiol. 110:1388 –1393. Virtanen, J., Parkkonen, L., Ilmoniemi, R. J., Pekkonen, E., and Na¨a¨ta¨nen, R. 1997. Biopotential amplifier for simultaneous operation with biomagnetic instruments. Med. Biol. Eng. Comp. 35: 402– 408. Virtanen, J., Rinne, T., Ilmoniemi, R. J., and Na¨a¨ta¨nen, R. 1996. MEG-compatible multichannel EEG electrode array. Electroencephalogr. Clin. Neurophysiol. 99:568 –570. Zatorre, R. J., Evans, A. C., Meyer, E., and Gjedde, A. 1992. Lateralization of phonetic and pitch discrimination in speech processing. Science 256:846 – 849.